Font Size: a A A

Research Of Neuro-Fuzzy Modeling For Nonlinear Systems

Posted on:2005-09-21Degree:MasterType:Thesis
Country:ChinaCandidate:H ZhangFull Text:PDF
GTID:2168360122975319Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
Combining neural network with fuzzy systems , a fuzzy-neural network is proposed. It is the focus of information science technique and intelligent control field. It is a kind of intelligence information control system which can imitate human's fuzzy inference. It is provded that it have universal approximation property. So studying identification problem on nonlinear systems based on neural fuzzy technique has feasibility, and it has important value in theory and application.The paper combines artificial neural network, fuzzy inference system, system identification, and does a thorough research on fuzzy-neural network. The research results are as follows:In this paper ,we propose a new neuro-fuzzy systems with Laplace (probability density function) membership function, and proved its universal approximation property by using Weierstrass theorem . We get excellent modeling results for nonlinear systems by applying the new neuro-fuzzy model.A new kind of optimal selection cluster algorithm (OSCA) is presented in this paper, and a new hybrid algorithm is obtained by combining OSCA with least squares method and gradient algorithm .The algorithm can synchronously solve the identification problems of the new fuzzy neural network model's structure and parameters. The higher identification precision is gained by apply the new fuzzy neural network model.A lot of simulation experiments are completed for different nonlinear systems in this paper, compared with the other methods, the new model obtains more excellent results. Simulation results show that the proposed scheme is very effective.
Keywords/Search Tags:artificial neural networks, fuzzy logical, fuzzy neural network, fuzzy modeling, nonlinear systems
PDF Full Text Request
Related items